Positron Annihilation Spectroscopy and Small Angle Neutron Scattering Characterization of Nanostructural Features in High-Nickel Model Reactor Pressure Vessel Steels

Journal of Nuclear Materials 351 (2006) 197–208 www.elsevier.com/locate/jnucmat

Positron annihilation spectroscopy and small angle neutron scattering characterization of nanostructural features in high-nickel model reactor pressure vessel steels

Stephen C. Glade a, Brian D. Wirth a,*, G. Robert Odette b, P. Asoka-Kumar c

a Nuclear Engineering Department, University of California, Berkeley, CA 94720-1730, USA b University of California, Santa Barbara, CA, USA c Lawrence Livermore National Laboratory, Livermore, CA, USA

Abstract

Irradiation embrittlement in nuclear reactor pressure vessel steels results from the hardening by a high number density of nanometer scale features. In steels with more than 0.10% Cu, the dominant features are often Cu-rich precipitates typically alloyed with Mn, Ni and Si. At low-Cu and low-to-intermediate Ni levels, so-called matrix hardening features are believed to be vacancy-solute cluster complexes, or their remnants. However, Mn–Ni–Si rich precipitates, with Mn plus Ni contents greater than Cu, can form at high alloy Ni contents and are promoted at irradiation temperatures lower than the nominal 290 C. Even at very low-Cu levels, late blooming Mn–Ni–Si rich precipitates are a significant concern due to their potential to form large volume fractions of hardening features. Positron annihilation spectroscopy (PAS) and small angle neutron scattering neutron (SANS) measurements were used to characterize the fine-scale microstructure in split- melt A533B steels with varying Ni and Cu contents, irradiated at selected conditions from 270 to 310 C between 0.04 and 1.6 · 1023 nm2. The objective was to assess the character, composition and magnetic properties of Cu-rich precipitates, as well as to gain insight on the matrix features. The results suggest that the irradiated very low-Cu and intermediate Ni steel contains small vacancy-Mn–Ni–Si cluster complexes, but not large, well-formed and highly enriched Mn–Ni–Si phases. The hardening features in steels containing 0.2% and 0.4% Cu, and 0.8% and 1.6% Ni are consistent with well-formed, non-magnetic Cu–Ni–Mn precipitates. The precipitate number densities and volume fractions increase, while their sizes decrease, with increasing Ni and decreasing irradiation temperature. The precipitates evolve with fluence in stages of nucleation, growth and limited coarsening. 2006 Elsevier B.V. All rights reserved.

1. Introduction Si–Mo–Ni low-alloy bainitic steels. RPVs operate at around 290 C accumulating fast neutron Western reactor pressure vessel (RPV) steels are ﬂuences from about 1 to 10 · 1023 nm2 over a fabricated from quenched and tempered C–Mn– 40–60 year service life, corresponding to a maxi- mum damage dose less than 0.15 displacements * Corresponding author. Tel.: +1 510 642 5341; fax: +1 510 643 per atom (dpa) [1–4]. However, even this relatively 9685. low dose is suﬃcient to produce severe embrittle- E-mail address: [email protected] (B.D. Wirth). ment in some cases, characterized by upward shifts

0022-3115/$ - see front matter 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.jnucmat.2006.02.012 198 S.C. Glade et al. / Journal of Nuclear Materials 351 (2006) 197–208 in the transition temperature (DT) between the more and fluence. In general, the results derived from brittle cleavage and more ductile microvoid coales- the different techniques are in good agreement, cence fracture regimes. although some questions remain regarding the com- Embrittlement is primarily the result of irradia- position of the precipitates, particularly with respect tion hardening, reflected in yield strength increases, to their possible Fe content [13]. Specifically, some with corresponding DT values of 300 C or more. atom probe data has been interpreted to suggest Embrittlement is controlled by a complex combina- the precipitates contain up to 50% or more Fe, tion of variables, including the neutron flux, fluence which is inconsistent with the results of other tech- and energy spectrum, the irradiation temperature niques. It is believed that some Cu catalyzes MNP (irradiation variables), and the alloy’s starting formation, but they can also form even in steels with microstructure and composition (material variables) little or no Cu at sufficiently high fluence, depending [1–6]. Important compositional variables include Cu on the irradiation temperature and alloy Ni and Mn (0.02–0.5%), Ni (0.2–2%) and Mn (0.3–2%), while P content. Once nucleated, MNPs grow rapidly to (0.005–0.040%) and Si (0.2–0.7%) generally play large volume fractions, with a correspondingly high secondary roles (note all compositions in this paper level of hardening and embrittlement [1,2]. How- are given in weight percent unless otherwise noted). ever, a complete understanding of the necessary The principal source of irradiation hardening is the conditions for MNP formation and evolution as a formation of high number densities of nanometer- function of irradiation and material variables does sized irradiation-induced precipitates and so-called not yet exist. matrix features, which impede the motion of dislo- In this paper, we present the results of a positron cations [1–6]. In Cu-bearing steels, irradiation hard- annihilation spectroscopy (PAS) and small angle ening is primarily associated with the accelerated neutron scattering (SANS) study to characterize formation of nanometer-sized Cu-rich precipitates the nanometer precipitates and matrix features (CRPs) due to radiation-enhanced solute diffusion. formed in a series of model RPV steels with con- The CRPs are enriched in Ni, Mn and Si. Mn–Ni trolled variations in Cu and Ni contents following rich precipitates (MNPs) containing a range of selected neutron irradiation at a flux of 3.2 ± Cu, as dictated by the alloy content of this element, 0.6 · 1015 nm2 s1 to fluences from 0.04 to 1.6 · form at high alloy Mn and Ni levels, and are pro- 1023 nm2 over the temperature range of 270– moted by lower irradiation temperatures (<290 C) 310 C. The results, which extend our previous mea- and Cu contents [2,7]. Indeed, nearly pure well- surements on model alloys, provide quantitative formed Mn–Ni–Si phases develop even in very information on the composition, sizes, number low-Cu steels at sufficiently high fluence and Ni densities, volume fractions and magnetic properties contents (possibly > 1%), although the tempera- of the Cu–Ni–Mn precipitates (CRPs), as well as ture–alloy composition MNP phase boundaries further qualitative insight on the matrix features. have not been experimentally established. Smaller, so-called matrix features, believed to be vacancy- 2. Experimental methods solute (e.g., Mn, Ni and Si) cluster complexes, or their solute remnants, also contribute to hardening, 2.1. Materials and irradiations even in alloys with low or no copper [3,8]. The matrix features may be precursors to MNP Commercial model (CM) A533B-type split-melt formation. bainitic steels with controlled variations in Cu and The CRPs [9] and MNPs [1], first predicted by Ni content, as shown in Table 1, were examined in Odette, have been identified and characterized by this study. The matrix consisted of a low-Cu, inter- a combination of experimental techniques, includ- mediate Ni steel (CM3) and intermediate (CM16 ing small angle neutron scattering (SANS) [2,10– and CM17) and high (CM19 and CM20) Cu alloys 13], atom probe field ion microscopy (APFIM) with both intermediate (CM16 and CM19) and high [13–19], combined electrical resistance Seebeck coef- (CM17 and CM20) Ni contents. These CM-series ficient measurements (RSC) [19–21], and recently, steels had typical prior austenite grain sizes of positron annihilation spectroscopy (PAS) [22–26]. 50 lm and tempered bainite microstructures. The The characteristics of the CRPs and MNPs have heat treatment schedule was as follows: austenitize been studied as a function of steel composition, heat at 900 C for 0.5 h, salt bath quench to 450 C treatment, irradiation temperature, neutron flux and hold for 600 s, temper at 660 C for 4 h, stress S.C. Glade et al. / Journal of Nuclear Materials 351 (2006) 197–208 199

Table 1 Measured compositions of the model steels used in this study Steel Cu (wt%) Ni (wt%) Mn (wt%) Mo (wt%) C (wt%) P (wt%) Si (wt%) S (wt%) CM3 0.02 (low) 0.85 (intermediate) 1.60 0.49 0.13 0.006 0.16 0.000 CM16 0.22 (intermediate) 0.82 (intermediate) 1.58 0.51 0.16 0.004 0.25 0.000 CM17 0.22 (intermediate) 1.59 (high) 1.54 0.50 0.16 0.004 0.25 0.000 CM19 0.42 (high) 0.85 (intermediate) 1.63 0.51 0.16 0.005 0.16 0.003 CM20 0.43 (high) 1.69 (high) 1.63 0.50 0.16 0.006 0.16 0.003 All values are given in weight percent. relieve at 607 C for 24 h, cool at 8 C/h to 300 C. (or lower) positron affinity than the matrix they Air cooling followed all the steps in the heat treat- are embedded in, also leading to positron localiza- ment sequence described above. tion in these regions. In this study of model Approximately 1 · 1 · 0.2 cm coupons were A533B-type steels, the positron affinities of the neutron irradiated in the University of California, key elements are: 3.72 eV Mn, 3.84 eV Fe, Santa Barbara (UCSB) irradiation variable (IVAR) 4.46 eV Ni and 4.81 eV Cu [29]. As shown by experiment at the University of Michigan Ford Nagai et al. [22], PAS is an especially well suited Nuclear Reactor. The neutron fluxes, fluences and technique for studying CRPs, since Cu clusters as irradiation temperatures are shown in Table 2. small as 0.6 nm embedded in an Fe matrix effec- The irradiations provided a systematic variation in tively localize and strongly trap positrons. fluence from 4.0 · 1021 to 1.6 · 1023 nm2 (T11, Ultimately, the positron annihilates with an elec- T12, T14, and T16) at a nominal flux of 3 · tron, predominately yielding two 511 keV photons 1015 nm2 s1 and 290 C, as well as a systematic traveling in approximately opposite directions. The variation in irradiation temperature from 270 to photons carry information about the positron 310 C (T16, T18, and T20) at the same flux and a annihilation site. The momentum of the electron fluence of 1.6 · 1023 nm2. (primarily)-positron (thermalized) pair prior to annihilation produces a Doppler shift (blue-shift 2.2. Positron annihilation spectroscopy (PAS) and a red-shift) of the two annihilation photons. The energy shift of each photon is given by PAS is a well-established technique for detecting 1 1 2 open-volume regions in a material [27], as well as DE ¼ p c ¼ hLm0c . ð1Þ 2 L 2 the chemical identity of the elements at the annihilation site [28]. After they thermalize, positrons either Here DE is the photon energy difference from the annihilate in the matrix or localize in open-volume nominal value, pL is the corresponding longitudinal regions (vacancies, nanovoid clusters, and other momentum shift along the direction of the gamma defects with regions of dilatational strain) due to ray emission, hL is the angular deviation of the pho- the ‘absence’ of positively charged atomic nuclei. tons from 180, m0 is the electron rest mass, and c is Such trapped positrons also have a longer lifetime the speed of light. The pL is expressed here in atomic due to the lower electron density in the open-volume units (1 a.u. = 7.28 mrad · m0c). The coincidence defects. Since each element has a unique positron Doppler broadening (CDB) technique is used to affinity, second phase precipitates may have higher evaluate DE, hence pL, by simultaneously measuring the two photon energies. Since the pL have element specific spectral distributions that correspond to the Table 2 Neutron irradiation conditions used in this study momentum distributions of the annihilation electrons, analysis of the orbital electron momentum Neutron flux Neutron fluence T (C) [/ (n m2 s1)] [/ (n m2)] spectrum (OEMS) provides information on the local composition at the annihilation site, including T11 2.6 · 1015 4.0 · 1021 290 15 22 those in strong positron traps. The OEMS is typi- T12 3.2 · 10 1.0 · 10 290 T14 3.2 · 1015 4.8 · 1022 290 cally represented as the fraction of annihilations T16 3.0 · 1015 1.6 · 1023 290 (relative number of counts) at each momentum 15 23 T18 3.6 · 10 1.7 · 10 270 interval as a function of pL, n(pL), normalized by a 15 23 T20 3.4 · 10 1.6 · 10 310 standard spectrum, in our case that for nominally 200 S.C. Glade et al. / Journal of Nuclear Materials 351 (2006) 197–208 pure and defect free Fe, nFe(pL), or n(pL)/nFe(pL). In and transported to the specimen using a strong transition metals in the Fe-series, lattice annihila- magnetic field (1.0 kG). The magnetic field can tions preferentially occur with 3d orbital electrons. also be used to perform so-called spin-polarized, The maxima and minima of the n(pL)/nFe(pL) magnetic positron annihilation measurements [24], OEMS are specific to a given element. For example, yielding data on the magnetic character at the the normalized OEMS of Cu and Ni have peaks at positron annihilation site. The magnetic field is an intermediate position in the high-momentum re- alternately oriented parallel and anti-parallel to gion. In contrast, other elements, such as Mn, have the positron spin. Changes in the majority and a minimum in the Fe-normalized OEMS at high minority electron populations in a ferromagnetic momentum. Since annihilations in vacancies and material with magnetic field reversal, combined with vacancy cluster traps occur primarily with the delo- the fact that electron–positron annihilation only calized valence electrons of the surrounding atoms, produces two annihilation photons when the respec- the corresponding OEMS are characterized by the tive spins are anti-parallel, results in corresponding enhancement of the low-momentum region. differences in the OEMS measured by CDB. When Mixtures of elements and vacancies have OEMS the OEMS data are represented in S–W plots, the that are approximately the constituent fraction two data points, associated with each magnetic field based averages, weighted by factors that reflect their orientation, are split. Non-magnetic annihilation relative affinities. The weighting can be determined sites have almost identical OEMS and S–W plot from first principles quantum mechanical calcula- positions for the parallel and anti-parallel magnetic tions. Measured OEMS also reflect a competition field orientations, while annihilations at atoms with between various annihilation sites, including the different majority–minority electron spins show a matrix and various higher affinity traps. Positrons characteristic S–W splitting. The observed degree generally do not detect low affinity features. Finally, of splitting is again a function of the weighting of positrons may either localize within a sub-region of the various constituents in an annihilation site and a compositionally non-uniform precipitate or between various trapping sites. defect-solute complex, or the positron wave function may be incompletely confined (leak out into 2.3. Small-angle neutron scattering (SANS) the surrounding matrix) of very small trapping features. For example, positrons would tend to be SANS was performed at the 8-m (NG1) beamline localized in nearly pure Cu-rich core of a CRP, at the Cold Neutron Research Facility of the hence not fully detect enrichment of Mn, Ni and National Institute of Standards and Technology Si atoms in a surrounding shell. In both cases, the [30]. The un-irradiated (control) materials and the overall OEMS is a spatially and affinity weighted irradiated samples were each measured in a strong combination of the constituents comprising the magnetic field (P1.8 T) oriented horizontally to sat- feature. urate the a-Fe matrix. This allows for measurement In addition to n(pL)/nFe(pL) versus pL plots, it is of both the nuclear (N) and magnetic (M) neutron convenient to represent OEMS data in so-called cross-sections as a function of the scattering vector, S–W plots. The S is a fraction of low-momentum q, which results from the corresponding nuclear and annihilations defined by specified pL limits, which magnetic scattering length density differences in our case is pL 6 0.382 a.u., and W is the corre- between the a-Fe matrix and the feature. The sponding high-momentum annihilation fraction as neutron scattering at small angles (h 6 8) was mea- specified by a corresponding second set of pL limits, sured from a well-collimated beam of cold neutrons here 1.0 6 pL 6 4.0 a.u. A particular microstructure (k 0.5 nm) with a two-dimensional 64 · 64 cm produces a point on a S–W plot. The weighting of position sensitive detector positioned 2 m from two elements (or annihilation sites, including the sample, and rotated 5 off-axis to increase the defects) is shown by the position of the specimen measured q-range. point on a line joining the pure elements (and/or After correcting the measured SANS data for sites). background and parasitic scattering, the small angle The CDB measurements were performed at Law- scattering produced by the irradiation-induced fea- rence Livermore National Laboratory (LLNL) tures was isolated by subtracting the corresponding using an experimental setup in which positrons scattering signal of the appropriate unirradiated emitted from a 22Na radioactive source are confined (control) specimen from that of the irradiated spec- S.C. Glade et al. / Journal of Nuclear Materials 351 (2006) 197–208 201 imen. The resulting irradiation-induced feature CM16 (0.22% Cu, 0.82% Ni, 1.58% Mn) 2.2 scattering was converted to an absolute scattering T11, 4.0E21 n m-2 cross-section using the measured scattering from 2.0 T12, 1.0E22 n m-2 an isotropic scattering water specimen with a known 1.8 T14, 4.8E22 n m-2 1 1 cross-section (= 0.88 cm ster ). T16, 1.6E23 n m-2 1.6 ) Cu

The nanometer Cu–Ni–Mn precipitates are L (p assumed to be roughly spherical and non-magnetic 1.4 Ni Fe Mn

in a saturated ferromagnetic iron matrix, providing )/n

L 1.2 a known magnetic scattering contrast. A log-normal n(p size distribution is fit to the absolute scattering 1.0 cross-section data as a function of q. The slope cur- 0.8 vature of the scattering curves determines the mean 0.6 size (hri) and size distribution of the features, while the absolute scattering intensity determines the 0.4 012345 corresponding number density (Nd) and volume pL (a.u.) fraction (fv). Further, the measured magnetic to nuclear scattering ratio (M/N) provides information Fig. 1. Normalized OEMS of the intermediate-Cu, intermediate- on the scattering feature composition. Additional Ni steel (CM16) under irradiations with increasing fluence from details regarding the SANS experiments and data 4.0 · 1021 to 1.6 · 1023 nm2 (T11, T12, T14, and T16). The analysis can be found elsewhere [13,31–33]. correspondingly normalized curves for elemental Cu, Ni, and Mn are shown for comparison. 3. Results composition from SANS M/N ratios [13] and resis- 3.1. PAS results tivity Seebeck coefficient measurements [19], as well as thermodynamic calculations [1,4]. However, due Fig. 1 presents the n(pL)/nFe(pL) OEMS obtained to the elemental affinity differences, the annihila- as a function of increasing fluence from 4.0 · 1021 to tions do not scale in direct proportion to the atomic 1.6 · 1023 nm2 for the model steel containing fractions. No increase in the low-momentum OEMS intermediate concentrations of Cu and Ni (CM16 region, indicative of positron trapping and annihila- with 0.22% Cu, 0.82% Ni) irradiated at 290 C. tion in vacancies and vacancy clusters, is observed. The normalized OEMS for the Fe reference speci- The OEMS trends shown in Fig. 1 are broadly con- men would correspond to a horizontal line at 1.0. sistent with those observed in other Fe–Cu–Ni–Mn Fig. 1 also shows the corresponding normalized model alloys and steels [25,31,35], including the OEMS of well-annealed elemental Cu, Ni, and intermediate-Cu, high-Ni steel (CM17, 0.22% Cu, Mn. The normalized OEMS for the irradiated steels 1.59% Ni) measured in this study, but not shown. show a broad peak at pL 3.25 a.u., consistent with Fig. 2 presents the normalized OEMS for all the positron localization in features containing Cu and CM steels as a function of increasing irradiation Ni. The magnitude of the peak at pL 3.25 a.u. temperature from 270 to 310 Cat1.6 · increases with increasing fluence up to a value about 1023 nm2. In all cases, the high-momentum OEMS 90% of the average Cu–Ni peak height at a fluence peak increases significantly between 270 and 290 C of 4.8 · 1022 nm2, and then remains approxi- and slightly between 290 and 310 C. The OEMS mately constant at higher fluence. It is well known peak also shows slight, but systematic, increases that there is Mn (and Si) in the precipitates [1– with higher Cu and decreases with higher Ni. Both 3,13,15,34]. Thus the results can be interpreted as the temperature and composition trends are consis- follows. The OEMS peak initially increases with tent with expected changes in the precipitate compo- fluence due to the increase in the size and number sition: higher temperature and Cu results in smaller density of precipitate trapping sites and correspond- quantities of Mn and Ni in the CRPs, while higher ing decrease in the competing annihilations in the Ni decreases the Cu content of the precipitates, Fe matrix. At higher fluence, the peak OEMS values resulting in MNPs versus CRPs. Note the similari- are roughly consistent with 70% of the annihilations ties between the alloys with intermediate and high with Cu and 15% with each Ni and Mn. These val- bulk Cu levels are partly due to the fact that the ues are reasonably close to estimates of the CRP actual Cu in solution prior to irradiation only varies 202 S.C. Glade et al. / Journal of Nuclear Materials 351 (2006) 197–208

Fig. 2. Normalized OEMS of all steels in this study [(a) CM3, (b) CM16, (c) CM17, (d) CM19, and (e) CM20] under irradiation conditions with variations in temperature (270 C – T18, 290 C – T16, and 310 C – T20). The nominal neutron fluence of these irradiations was 1.6 · 1023 nm2. from 0.22% versus 0.25% to 0.31%, respectively. in the trapping feature. However, the (oversimpli- Similar, but somewhat larger differences in the fied) assumption of preferential annihilations with OEMS peak would be expected based on the actual Cu and Ni versus Mn qualitatively rationalize the precipitate composition trends with changes in tem- somewhat lower compositional and temperature perature and alloy solute content, assuming the sensitivity of the OEMS peak. The other notable annihilations scale directly with the atomic fractions feature in the OEMS for the Cu-bearing alloys is S.C. Glade et al. / Journal of Nuclear Materials 351 (2006) 197–208 203 that there is a slight enhancement of the low- CM3 (0.02% Cu, 0.85% Ni,1.60% Mn) momentum region for the 270 C irradiation. As 0.23 discussed below this could be due to enhanced anni- 0.22 hilations with vacancies, Mn or some combination Ni 0.21 Cu of both. This observation is consistent with the expectation that the fractions of both of these con- 0.20 stituents are expected to increase in the irradia- un-irradiated 0.19 control tion-induced features at lower temperature. Fe 0.18 T20 In contrast to the behavior of the Cu-bearing T18 T16 alloys, the normalized OEMS of the very low-Cu 0.17

steel (CM3 with 0.02% Cu and 0.85% Ni) has a high momentum fraction 0.16 shallow minimum at higher momentum and a small enhancement at low momentum. This OEMS behav- 0.15 low-Cu, intermediate-Ni Mn ior is consistent with positron annihilation in some 0.14 combination of vacancies, small vacancy clusters, 0.38 0.39 0.40 0.41 0.42 0.43 0.44 and vacancy-solute clusters containing Ni, Si and a low momentum fraction Mn. While the positron affinity is low for Mn, it CM19 (0.42% Cu, 0.85%Ni, 1.63% Mn) seems likely that positrons that do not annihilate 0.23 in the matrix or at defects, such as dislocations and 0.22 interfaces, would be trapped by very small vacancy Ni Cu cluster-solute complexes. Note, that positron inter- 0.21 T20 T16 action and annihilation with Si atoms will be analo- 0.20 gous to that with a vacancy due to a combination of T18 high affinity (6.95 eV) [27] and an OEMS curve 0.19 Fe similar to Mn. Indeed at high momentum the 0.18 un-irradiated n(pL)/n(pL)Fe (<0.5) for Si is lower than that for control 0.17 Mn while Si has a slightly larger (0.25) peak at high momentum fraction low momentum [36]. Thus, in the case of the 0.16 low-Cu CM3 alloys, the OEMS likely reflects some 0.15 Mn weighting of a mixture of vacancies, Ni, Mn, Si high-Cu, intermediate-Ni and Fe as well as a balance between annihilations 0.14 0.38 0.39 0.40 0.41 0.42 0.43 0.44 in the matrix and small trapping features. b low momentum fraction The results of spin-polarized, magnetic positron annihilation measurements are presented in the S– Fig. 3. Results of the spin-polarized, magnetic positron annihilation measurements for the (a) low-Cu, intermediate-Ni (CM3) W plots in Fig. 3. Ferromagnetic Fe and Ni both and (b) high-Cu, intermediate-Ni (CM19) steels. Positron anni- exhibit splitting between the measurements with hilation fraction in the high-momentum region (1 a.u. 6 pL 6 the magnetic field oriented parallel (up-triangle) 4 a.u.) versus the low-momentum region (pL 6 0.382 a.u), nor- and anti-parallel (down-triangle) relative to the pos- malized to the total annihilations, for the magnetic field oriented itron spin, while the points for non-magnetic Cu parallel (up-triangle) and anti-parallel (down-triangle) to the positron polarization. Elemental Fe, Cu, Ni, and Mn are shown and Mn fully overlap. As expected, the un-irradi- with solid symbols. The line in (a) is drawn to guide the eye. ated steels exhibit a split similar to that of elemental Fe. The splitting in irradiated low-Cu, intermediate- Ni steel (CM3) shown in Fig. 3(a) is similar to that and bigger low-momentum fraction. The S–W data in the unirradiated alloy. However, the data falls on points move further in the direction of Mn (and/or the line roughly connecting Ni and Mn, and passing Si and vacancies) with decreasing irradiation close to points for the unirradiated steel. The posi- temperature. tion of the data points between the unirradiated There is no unique interpretation that can be steel and Mn suggests enhanced annihilations at applied to these results. However, they are broadly the latter. Note the point for vacancy features is consistent with some positron annihilation at very not shown, but lies close to Mn on the S–W plot. small vacancy-Mn–Si–Ni–Fe complexes. The The corresponding point for Si falls off the plot in balance of annihilations occurs in the Fe-matrix. the direction of a smaller high-momentum fraction The strongest evidence of the involvement or 204 S.C. Glade et al. / Journal of Nuclear Materials 351 (2006) 197–208 irradiation-induced features in the shifts in the The splitting observed in the S–W plot in unirra- direction of Mn, Si and vacancies, and that the diated high-Cu, intermediate-Ni steel (CM19) vacancy and/or Mn-Si content would be expected shown in Fig. 3(b), is smaller than that in the unir- to increase with decreasing temperature. However, radiated low-Cu, intermediate-Ni steel (CM3) in lifetime measurements are needed to determine the Fig. 3(a). This may be due to positron trapping in relative involvement of vacancies versus Mn, Fe some Cu–Ni–Mn clusters formed during the heat and Ni. Nevertheless, assuming there is trapping treatment prior to irradiation. Similar behavior is taking place, vacancies and/or vacancy-solute clus- observed in other Cu-bearing steels (CM16, CM17 ters are likely to be present. It may also be possible and CM20). Following irradiation of the high-Cu, that small features composed of Ni–Si–Mn–Fe intermediate-Ni steel (CM19) at temperatures from clusters could both trap positrons and yield the 270 to 310 C to 1.6 · 1023 nm2, the spin-polarized observed S–W data point trends. The splitting can data points fully overlap and do not show any split- be explained by the presence of both Ni and Fe in ting. This indicates that essentially all the annihila- the dilute clusters and their very small size, resulting tions occur at trapping sites with non-magnetic in incomplete positron localization in the feature atoms. The position of the points on the S–W plot itself, as well as competing annihilations in the Fe depends on the irradiation temperature. At 290 matrix. and 310 C, the points indicate features highly

Fig. 4. Results of SANS data analysis for the intermediate-Cu, intermediate-Ni (CM16) and the intermediate-Cu, high-Ni (CM17) steels with increasing fluence: (a) mean radius (hri), (b) number density (Np), (c) volume fraction of scattering features (fp), and (d) magnetic to nuclear (M/N) scattering ratio. The intermediate-Cu, intermediate-Ni steel (CM16) at the 1.0 · 1022 nm2 irradiation fluence (T12) was not measured. All irradiations were performed at 290 C, with flux variations as indicated in the figure and in Table 1. The lines are drawn to guide the eye. S.C. Glade et al. / Journal of Nuclear Materials 351 (2006) 197–208 205 enriched in Cu and Ni with a third element known • The fp exceed the atomic fraction of the initially to be Mn. The S–W points for the 270 C irradia- dissolved Cu content of the alloys, estimated to tion fall on or near a line connecting Cu and a larger be 0.19% and 0.22–0.29% for the intermediate amount of Mn (or alternately Si or vacancies or and high-Cu steels, respectively (Fig. 5(c)). vacancy clusters), although this does not preclude • The M/N increases with increasing irradiation the presence of Ni as well. The S–W plots for the temperature and decreasing alloy Ni content, other Cu-bearing CM-series steels (CM16, CM17 consistent with increasing precipitate Cu content and CM20) are very similar to the results shown (Fig. 5(d)). in Fig. 3(b). These results are generally consistent with previ- 3.2. SANS results ous observations [32,33].

Fig. 4 summarizes the SANS data for the inter- 4. Discussion and summary mediate-Cu steels with intermediate (CM16 with 0.22% Cu, 0.82% Ni) and high Ni (CM17 with Both the PAS and SANS measurements show the 0.22% Cu and 1.59% Ni) contents irradiated at formation and growth of Cu–Ni–Mn precipitates in 290 C over a range of fluences. Fig. 4(a) shows that Cu-bearing steels. First consider the high fluence the mean precipitate radii, hri, increases with data at 290 C. Increasing the alloy Ni content from increasing fluence, and are smaller for the high Ni about 0.8% to 1.6% increases fp and refines the (CM17) alloy. Fig. 4(b) shows the precipitate num- precipitate distribution with an increased Np. The ber density, Np, is systematically higher in the high increase in fp is associated with higher Mn and Ni Ni (CM17) steel, with a peak at intermediate content of the precipitates as reflected in lower M/ fluence. The Np in the intermediate Ni (CM16) steel N ratios of 1.45 ± 0.05. This nominal M/N results increases slowly and monotonically with increasing from a precipitate composed of approximately fluence. Note in all cases, the Np are high with val- equal quantities of Cu, Ni and Mn. Assuming that ues in excess of 1023 m3. As shown in Fig. 4(c), the the Cu is nearly fully precipitated, this composition precipitate volume fraction, fp, increases continu- estimate is also compatible with the measured ously with fluence. The fp is significantly larger in volume fractions in the high-Ni alloys (CM17 and the high-Ni (CM17) versus intermediate-Ni CM20); the high-Ni steels with intermediate and (CM16) steel at higher fluence, but are similar at high-Cu initially contain 0.19% and 0.28% atom low fluence. Fig. 4(d) shows that the magnetic to fraction of dissolved Cu, respectively, which are nuclear scattering ratios, M/N, initially increase, 28% and 33% of the corresponding total fp of but are approximately constant at fluences above 0.68% and 0.86%. A similar analysis of the inter- 1022 nm2. Note the SANS data are less reliable mediate-Ni steels (CM16 and CM19) yields consis- at the lowest fluence for the features due to their tent precipitate compositions of 44% Cu and small size hrpi and fp. The M/N are generally lower 28% of each Ni and Mn. Notably, in all these cases, in the high-Ni steel (CM17), indicating that increas- the irradiation-induced features are MNPs rather ing the alloy Ni content increases the concentration than CRPs. Finally, the precipitate Mn and Ni of both Ni and Mn in the precipitates, consistent contents increase with decreasing irradiation tem- with other measurements [1,2,32,33]. perature and vice versa. We have not considered Fig. 5 shows the corresponding plots of the Si, which would produce scattering contrast similar SANS data for all the Cu-bearing steels as a func- to a mixture of 2/3Mn and 1/3Ni. However, these tion of irradiation temperature from 270 to alloys have low Si content and the amount of Si 310 C. The results can be summarized as follows: present in the precipitates is expected to be modest. At first glance, the lower sensitivity of the OEMS • The precipitate hrpi increase and Np decrease to composition and temperature variations may with increasing temperature (Fig. 5(a) and (b)). seem puzzling. However, this behavior can be read- • The coarsening of the precipitate structure with ily rationalized by noting the competing effects of increasing temperature is generally accompanied Mn (and Si) versus Ni on the magnitude of the by modest reductions in fp (Fig. 5(c)). OEMS peak along with preferential annihilations • The precipitate hrpi decreases, while the Np and fp in Cu and to a lesser extent Ni versus Mn due to increase with increasing Ni (Fig. 5(a)–(c)). different elemental affinities, as well as precipitate 206 S.C. Glade et al. / Journal of Nuclear Materials 351 (2006) 197–208

Fig. 5. Results of SANS data analysis for the intermediate/high-Cu and intermediate/high-Ni steels (CM16, CM17, CM19, and CM20) with varying temperature: (a) mean radius (hri), (b) number density (Np), (c) volume fraction of scattering features (fp), and (d) magnetic to nuclear (M/N) scattering ratio. Some data points in Fig. 5(a) and (d) superimpose each other. The intermediate-Cu, intermediate-Ni steel (CM16) at the 310 C irradiation condition (T20) was not measured. The average fluence of the irradiations is 1.6 · 1023 nm2. See Table 1 for specific irradiation conditions. The lines are drawn to guide the eye. structures consisting of almost fully Cu-rich cores of the data on Cu-bearing steels could be discussed, surrounded by Ni, Mn and Si enriched shells. The in summary, all of the SANS and PAS observations broad consistency of the OEMS and SANS results are highly consistent with previous results based on is encouraging. However, a more important obser- a variety of characterization techniques in addition vation is the complete lack of magnetic splitting in to existing theoretical understanding and models the precipitate trapping features observed in the of the evolution of precipitate structures in irradi- spin-polarized positron measurements. This obser- ated RPV steels [1–4,6,12,13,19,24–26]. The one vation indicates the absence of significant quantities key exception to this overall consistency, is the of Fe and Ni in the annihilation region. However, apparent presence of Fe within the precipitate cores, this does not mean that the features do not contain as found in atom probe tomography studies. Ni along with Mn and even some small amount of The OEMS of the low-Cu, intermediate-Ni steel Fe in the shell region. (CM3) exhibit a weak signal probably associated There is an open question about Mn and Si ver- with some annihilations at small vacancy-Mn–Ni– sus vacancy source of the enhancement of the low- Si-Fe complexes. The presence of vacancies, Si and momentum region in the 270 C irradiations that Mn are qualitatively consistent with the OEMS can only be resolved by lifetime measurements that signature, while vacancies, Si and Ni would make will be carried out in the future. While other details the feature an attractive trapping site. The corre- S.C. Glade et al. / Journal of Nuclear Materials 351 (2006) 197–208 207 sponding SANS data was generally too weak and References scattered to fit reliably. The strongest scattering was observed in the 270 C irradiation to the highest [1] G.R. OdetteMaterials Research Society Symposium Pro- fluence, which indicates the presence of very small ceedings, vol. 373, Materials Research Society, Warrendale, Pennsylvania, 1995, p. 137. volume fraction with f 0.028% (nominal value [2] G.R. Odette, G.E. Lucas, Radiat. Eff. Def. Solids 144 (1998) assuming that the feature is non-magnetic) of extre- 189. mely small features, with hri0.5 nm and charac- [3] G.R. Odette, G.E. Lucas, JOM 53 (2001) 18. terized by a low M/N ratio of 1. Combined [4] G.R. Odette, Neutron Irradiation Effects in Reactor Pressure resistivity and Seebeck coefficients are consistent Vessel Steels and Weldments, International Atomic Energy Agency, Vienna, IAEA IWG-LMNPP-98/3, 1998, p. 438. with a volume fraction, f 0.17% of Ni–Mn–Si [5] G.R. Odette, B.D. Wirth, D.J. Bacon, N.M. Ghoneim, MRS clusters [19], not including any associated vacancies. Bull. 26 (2001) 176. The SANS and OEMS data are roughly consistent [6] G.R. Odette, T. Yamamoto, D. Klingensmith, Philos. Mag. with a very high density 4 · 1024 m3 of small 85 (2005) 779. 0.5 nm features, that remain partially magnetic [7] G.S. Was, M. Hash, G.R. Odette, Philos. Mag. 85 (2005) 703. (50% of the ferrite matrix), composed of 50% [8] G.R. Odette, B.D. Wirth, J. Nucl. Mater. 251 (1997) 157. Fe with the balance split between Ni, Mn and Si [9] G.R. Odette, Scripta Metall. 11 (1983) 1183. (probably a higher fraction of Ni) plus one to a [10] A. Ulbricht, J. Boehment, P. Strunz, C. Dewhurst, M.H. few vacancies. Such features would be expected to: Mathon, Appl. Phys. A – Mater. Sci. Process. 74 (2002) (1) manifest only weak magnetic and nuclear small S1128. [11] M. Grosse, J. Boehment, R. Gilles, J. Nucl. Mater. 254 angle neutron scattering; (2) trap positrons and (1998) 143. yield a weighted OEMS with a small minimum at [12] B.D. Wirth, G.R. Odette, P. Asoka-Kumar, R.H. Howell, higher momentum and enhancement at low momen- P.A. Sterne, in: G.S. Was (Ed.), Proceedings of the 10th tum that manifest magnetic splitting; and (3) International Symposium on Environmental Degradation of roughly account for the amount of solute removal Materials in Light Water Reactors, National Association of Corrosion Engineers, 2002. from the matrix as estimated from the RSC [13] M.K. Miller, B.D. Wirth, G.R. Odette, Mater. Sci. Eng. A measurements. Notably, there is no evidence of 353 (2003) 133. the development of well-formed, highly enriched [14] P. Pareige, K.F. Russel, R.E. Stoller, M.K. Miller, J. Nucl. MNPs in the low-Cu, intermediate-Ni steels (CM3). Mater. 250 (1997) 176. Positron lifetime measurements will be needed to [15] M.K. Miller, P. Pareige, M.G. Burke, Mater. Charact. 44 (1– 2) (2000) 235. further assess the validity of the hypothesis about [16] B. Radiguet, P. Pareige, A. Barbu, Sur. Int. Anal. 35 (2004) the character of matrix features outlined above. 515. However, it is noted that these conclusions are [17] M.K. Miller, M.G. Burke, J. Nucl. Mater. 195 (1992) 68. generally consistent with the combined OEMS and [18] P. Auger, P. Pareige, S. Welzel, J.C. Van Duysen, J. Nucl. lifetime observations of small vacancy cluster Mater. 280 (2000) 331. [19] G.R. Odette, C. Cowan, in: G.S. Was (Ed.), Proceedings of complexes in simple Fe–1%Mn alloys irradiated at the 10th International Symposium on Environmental Deg- 290 C [12,31]. radation of Materials in Light Water Reactors, National Association of Corrosion Engineers, 2002. Acknowledgements [20] S. Ishino, Y. Chimi, Bagiyono, T. Tobita, N. Ishikawa, M. Suzuki, A. Iwase, J. Nucl. Mater. 323 (2003) 354. [21] B. Acosta, F. Sevini, L. Debarberis, Int. J. Press. Vess. This work was partially supported by the US Piping 82 (2005) 69. Nuclear Regulatory Commission under contracts [22] Y. Nagai, M. Hasegawa, Z. Tang, A. Hempel, K. Yubuta, T. #04-94-049 and 04-01-064, and partially performed Shimamura, Y. Kawazoe, A. Kawai, F. Kano, Phys. Rev. B under the auspices of the US Department of Energy 61 (10) (2000) 6574. by the University of California, Lawrence Liver- [23] Y. Nagai, Z. Tang, M. Hassegawa, T. Kanai, M. Saneyasu, Phys. Rev. B 63 (2001) 134110. more National Laboratory under Contract No. [24] P. Asoka-Kumar, B.D. Wirth, P.A. Sterne, G.D. Odette, W-7405-Eng-48. We acknowledge the support of Philos. Mag. Lett. 82 (2002) 609. the National Institute of Standards and Technol- [25] S.C. Glade, B.D. Wirth, G.R. Odette, P. Asoka-Kumar, P.A. ogy, Center for Neutron Research, in providing Sterne, R.H. Howell, Philos. Mag. 85 (2005) 629. the neutron research facilities used in this work, [26] M. Hasegawa, Z. Tang, Y. Nagai, T. Chiba, E. Kuramoto, M. Takenaka, Philos. Mag. 85 (4–7) (2005) 467. and the major contributions of D. Klingensmith [27] P. Hautoja¨rvi, C. Corbel, in: A. Dupasquier, A.P. Mills Jr. (UCSB). (Eds.), Positron Spectroscopy of Solids, Proceedings of the 208 S.C. Glade et al. / Journal of Nuclear Materials 351 (2006) 197–208

International School of Physics, Course CXXV, Italian Ph.D. Dissertation, University of California, Santa Barbara, Physical Society, Bologna, Italy, p. 491. 1998. [28] P. Asoka-Kumar, M. Alatalo, V.J. Ghosh, A.C. Kruseman, [33] E.V. Mader, Kinetics of Irradiation Embrittlement and the B. Nielsen, K.G. Lynn, Phys. Rev. Lett. 77 (1996) 2097. Post-Irradiation Annealing of Nuclear Reactor Pressure [29] M.J. Puska, R.M. Nieminen, Rev. Mod. Phys. 66 (1994) 841. Vessel Steels, Ph.D. Dissertation, University of California, [30] C.J. Glinka, J.M. Rowe, J.G. laRock, J. Appl. Crystallogr. Santa Barbara, 1995. 19 (1986) 427. [34] C.L. Liu, G.R. Odette, B.D. Wirth, G.E. Lucas, Mater. Sci. [31] B.D. Wirth, P. Asoka-Kumar, R.H. Howell, G.R. Odette, Eng. A 238 (1997) 202. P.A. SterneMaterials Research Society Symposium Proceed- [35] S.C. Glade, B.D. Wirth, P. Asoka-Kumar, P.A. Sterne, G.R. ings, vol. 650, Materials Research Society, Warrendale, Odette, Mater. Sci. Forum 445&446 (2004) 87. Pennsylvania, 2001, p. R6.5.1. [36] K. Dohi, N. Soneda, personal communication. [32] B.D. Wirth, On the Character of Nano-Scale Features in Reactor Pressure Vessel Steels Under Neutron Irradiation,